The 17th Annual Personalized Medicine Conference – Part IV

[00:00] I hate to be the one to bring the break to an end, but we only have a few more sessions, and then we get to the reception. Thank you.

[00:20] everybody. I'm Darryl Pritchard, the senior vice president at the Personalized Medicine Coalition in charge of the science policy portfolio. That's the evidence that we developed to show the value of personalized medicine and the personalized medicine approach. It's my privilege to introduce

[00:40] Next session is our distinguished speakers from Health Advances, a life science and health technology consultancy based in Boston. Before the break, we had two sessions back to back on developing diagnostics and in creating new personalized medicines.

[01:00] But a recurring theme that we've all been talking about all day is, and after you've had those sessions where they talk about some of the challenges in developing these new technologies, it might not seem it, but as Mike Polanyi said earlier today, implementation is the primary limitation.

[01:20] now to the personalized medicine approach. Science is no longer the primary limitation. We need to think about implementation. So the next two sessions are going to talk just about that. Key to implementing

[01:40] personalized medicine and the use of these new personalized medicine technologies is providing evidence of the value proposition, evidence that these personalized medicine technologies are better and more beneficial to patients in the healthcare system. But stakeholders that need this evidence require it to be real world evidence.

[02:00] evidence, practice-based evidence so that it makes clear to providers and to payers that personalized medicine is the way to go in delivering personalized medicine and implementing a personalized medicine approach. With this in mind, the Personalized Medicine Coalition and a steering committee of experts from our membership

[02:20] partnered with Health Advances, Health Technologies and Life Sciences Consultancy based in Boston with real thought leadership in personalized medicine. To conduct a project on implementing

[02:40] personalized medicine called improvements in clinical care associated with personalized medicine and precision oncology. The key to this project is that it shows that

[03:00] personalize medicine approach improves clinical care, both to patients through better experiences and better outcomes and benefits the health care system overall. For the first time, we're going to present the findings for this study here today.

[03:20] I'm delighted to introduce our co-speakers, our co-presenters. Gary Gustafson, who's a partner and managing director at Health Advances, leads the personalized medicine practice. Gary is a recognized thought leader in personalized medicine, focused on commercialization, value studies, and access, which was

[03:40] isn't in your bio, but is absolutely true. Access and providing evidence to stakeholders who control access to new novel technologies, primarily payers and providers. I've been working with Gary for a long time and really, I can't think of anybody who understands better how to make those messages and how to

[04:00] approach the access question. Joining him is Arushi Agarwal. Arushi joined Gary in 2011 at HealthAdvance and she's a partner and now co-leads to Personalized Medicine Practice. Arushi is an accomplished policy researcher that we've also been working with for some time on a number of projects and she's focused

[04:20] on companion diagnostic, stakeholder perspectives, and integration of personalized medicine into the healthcare system. So Gary and Arushi, thank you for leading our discussion on improvements in clinical care.

[04:40] be joining the stage with so many leaders from today. So thank you to PMC for the opportunity. I'm going to move very quick through this because we only have 15 minutes, but I will say that we really want to engage with the community. So if there is any interest in learning more about the framework, the model that we're going to introduce or the data that came out of it, please see myself, Darryl Aruchian.

[05:00] after the talk. I also do want to briefly thank the Project Steering Committee. This group of really engaged individuals were critical in guiding our decision making, helping us navigate data resources, etc. So we've heard, as Darryl mentioned, throughout this talk everybody's been mentioning it's no longer the science or the

[05:20] diagnostic, it's everything from the test to getting the ultimate targeted treatment to the patient that is in the way of broad adoption of personalized medicine. And so when the PMC came to us to partner on developing a framework, a tool to effectively and holistically measure personalized medicine adoption.

[05:40] We knew we had to think holistically, right? We knew we had to come up with a broad set of criteria beyond just simply, is a health system testing or not? And so we put together this study in 2019. One of the things that made it different was this multifactorial approach that we took to defining the adoption.

[06:00] of personalized medicine and again that went beyond simply just testing. We looked at data utilization, the frequency of testing, leadership, funding, etc. So I'll talk a little bit about that in a bit and and also we went more broadly beyond just simply the academic institution. So we went deep into the community to understand what was going on. Broad geography.

[06:20] broadened it beyond oncology. We've heard that today as a key theme as well. The major outcomes of this study, we presented it at this meeting in Boston in 2019, subsequently published the novel maturity model in the Journal of Personalized Medicine, and maybe more importantly we've discovered that this tool can serve as a way to

[06:40] almost define a personalized medicine adoption fingerprint for a given health institution that then many industry stakeholders can look at and dissect and think about, well, what are the key challenges at this institution and how can we devise solutions to hopefully move them along from a solution standpoint to bring in personalized medicine.

[07:00] forward. So that brings us to today where we're going to be talking about our latest work where very naturally Personalized Medicine Coalition said we've got this great tool, but we need to be able to connect it to improvements in quality of care. Otherwise, what good is the tool? It's designed to measure personalized medicine.

[07:20] adoption, but we have to bring it forward. So the work that Arushi is going to talk through today is going to show how we correlate that model, that score to actually improvements in care and value. I'll quickly talk through how this multifactorial solution was put together. So naturally, the

[07:40] collection of genomic data that was certainly critical for us to build as one of the key foundational criteria, but we went beyond that. So multi-omics, proteomics, epigenomics, metabolomics, what else is the institution looking at? What else beyond clinical data? So we looked at or we know that when an institution brings together clinical outcomes data, economic

[08:00] social determinants of health. When they link that to biomarker data, that's when true progress moves forward. So we wanted to give credit for that. Testing guidance and data accessibility. Is this simply ordered by a clinician manually, or is it integrated into the electronic health record and results delivered in that manner?

[08:20] localization of data, how is the data being used? Is it simply to identify an on-label therapeutic or do they go beyond to identify unlabeled or non-label off-label therapeutics, clinical trial enrollment, research, etc. Data sharing, is the data shared internally? Is it shared externally? Are they joining the ecosystem in that way? Is there internalization?

[08:40] personalized medicine leadership at a physician level, at a department level, or is there a true C-suite personalized medicine initiative commitment in the institution? And then finally, how is it funded? Are we largely looking to funding through natural reimbursement or is there internal funding that can really push forward some of these unique initiatives within the institution? So that was our

[09:00] broad-based multifactorial score that is developed from that. And then what we did was we surveyed 153 different institutions across the US and came up with five different levels of precision, personalized medicine adoption and scored them appropriately. And you can see a nice normal distribution across the different

[09:20] levels of scoring. And I think we'd all hope that if we reran this today, we'd see a natural shift to the right of this chart. But that brings up the key question, right? If we are expecting institutions to invest to improve their personalized medicine adoption, perhaps via this score, then we also have to prove that that will see some

[09:40] return in clinical improvement, in value, in patient care. And that's what Arushi, my colleague, is going to talk through now. Great. So as Gary mentioned,

[10:00] The goal of really developing this framework was to be able to use it as a fingerprint for assessing any institution and their level of personalized medicine integration. And so what we wanted to do was look at institutions where we could then get data on the back end to understand how they were actually implementing that personalized medicine. So thinking of it as the framework and the personalized medicine.

[10:20] medicine score, as the access, the resources, and the infrastructure to be able to deliver on the promise of personalized medicine, but then taking it one step further to actually saying, are you using that access and information and resources and infrastructure to do something that's meaningful and valuable to patients?

[10:40] kind of the metrics that we would use for value to patients as really thinking about are patients getting the best treatment options possible. And that could be a targeted therapy based on their tumor type and whether or not there are targeted therapies available and approved. Or it could mean a clinical trial where in some tumor types that's really the only option, but it is sometimes the

[11:00] best option or the best therapeutic option. We also wanted to look at this across a couple of different tumor types. One that we saw is more of a gold standard or established tumor type for precision medicine, so one cancer, and one where things were a little more emerging where there may be less options available but still sort of emerging hints of precision medicine where we could potentially still see some

[11:20] correlation between the score and actual care delivery. So we did look at six institutions from the SIAPs network to be able to get access to all of their data. A number of these institutions were clustered within the same health system, but they were individual institutions with individual practices, representing a diverse set of

[11:40] organizations across the country. So what we found was pretty interesting. Ultimately what we found was that the precision medicine scores are really the strongest indicator of care quality when there are many therapy options to be had and that could be many targeted therapy options, many clinical trial options.

[12:00] But really if we put a very simple point on it, if you have the infrastructure to act on the promise of precision or personalized medicine, if there's enough personalized medicine to be had, you can actually improve the quality of care. But if you have only limited options, if there are only a few therapies or one or two things that you can do, you could have all the resources.

[12:20] in the world, but that doesn't actually impact or improve outcomes. So sort of the corollary to all of this is there's a broader call to action and sort of motivation on the industry level to continue to develop and have these clinical trials and these personalized therapies so we can continue to have these options and allow folks to take advantage of all that.

[12:40] infrastructure that they have. But at the end of the day, what we found was that there's sort of a weaker score correlation, both on the targeted therapy side and the clinical trial side, when there weren't that many options to be had, a stronger correlation when there were, and the caveat being on the clinical trial side, as we all know, there's a number of factors that also influence and sort of confound the ability.

[13:00] to be able to be enrolled in a clinical trial. And so while we did see a correlation, it wasn't quite as strong as what we saw on the targeted therapy side. So we'll walk through a couple of examples. We did look at every single one of these indications, but wanted to highlight a few kind of notable things that we were noticing in the data. So at first glance, when we saw this data,

[13:20] I think we were kind of scratching our heads and wondering if this was good news or bad news. But at the end of the day, I think what we're seeing is this is a hallmark example of how we could get to a point within an indication where the personalized medicine integration score honestly doesn't matter. You could be doing all of the best things in the world or you could be doing nothing, but from an indication standpoint,

[13:40] standpoint, we've reached a point for HER2-positive breast cancer where institutions just understand that you're doing patients a disservice if you're not putting them on HER2-targeted therapies. And it doesn't matter what levels of integration that you have or don't have, that is just what you should be doing. And so not only was there not a correlation, you can see.

[14:00] that the rates of treatment are 100% are close to 100%, which is ultimately what we want to see, and I think sort of that future vision of every indication should hopefully look like this. But at the same time, the framework that we've put together is valuable and it is that valuable fingerprint as we were talking about and where it really does become valuable is when there are areas

[14:20] where there's more emerging options that exist and that PM integration score can actually be a really valuable tracker to see how institutions are doing today and potentially track their progress over time. So this is where we saw a nice correlation between higher precision medicine score and higher use of targeted therapies and this was from metastatic lung cancer which is an area where there's you know 20 approved targeted therapy.

[14:40] The list continues to grow. You know, that commitment to sophisticated precision medicine practices ultimately results in better care for patients and more patients getting on the treatments that they need to be getting on. On the clinical trial side, as we mentioned before, I think this is where we saw

[15:00] Slightly less strong correlation knowing, again, that there are some confounding factors that may influence whether or not a patient actually ends up getting on a clinical trial, even though the intention may be there, or the access or the information may be there. But at the end of the day, it was the same thing, where, you know, we saw clinical areas where there's a lot of clinical trial activity, metastatic lung cancer, where there's a number of

[15:20] number of trials for targeted therapies and then within breast cancer, TNBC has some highest activity of clinical trials as well and these are areas where we are seeing again greater commitment to precision medicine and greater commitment and infrastructure for integration of precision medicine actually leads to potentially better care delivery on that front as well.

[15:40] So as Darryl mentioned, we are excited about what we've been finding. We are looking forward to publishing our results in a paper similar to what we did with the last study. So keep an eye out for that. And then as Gary was mentioning too, I think this was a really interesting case study where we got to look at a handful of institutions, but what we'd love to

[16:00] continue to build on this and actually expand it beyond just the six institutions. Ideally get a broader set of institutions, potentially even look at some additional metrics and outcomes if possible. And so we'd love to chat with you all later in this meeting or offline to get your thoughts on how we might be able to do that as I think we do feel.

[16:20] something quite promising. So thank you all. Thank you again to the steering committee. I know many of you are in the audience and thank you again to Tom and the SIAPs team for all of your help with the data analysis and collaboration. We really appreciate it.

[16:40] Thank you, Arushi and Gary. Unfortunately, we don't have enough time to take a bunch of questions. We have to get into the next session, but I'm sure there are questions. So Gary and Arushi and myself are available, Gary and Arushi, right away. I'm going to introduce the next session, but really we encourage you all to follow up with both

[17:00] Health Advances and the Personalized Medicine Coalition about the study and the overall concept of demonstrating the clear value of a personalized medicine approach for improving clinical care. So thank you again, Arushi and Gary. The next session I'm just going to turn it over to Arushi.

[17:20] over to Howard McLeod to introduce, I'll introduce Howard, but session really focusing on clinical adoption of personalized medicine, progress and lessons learned. Unfortunately, we had planned on doing this with a US and a European perspective and our

[17:40] speaker from Europe, Tony Andrew, fell ill and couldn't make the trip from Europe. So we are probably going to be more US centric, but let me turn it over to Howard McLeod to introduce the panelists and to lead the discussion here on clinical adoption of personalized medicine challenges and strategies. Howard is

[18:00] The managing director at Clarified Personalized Medicine is also probably pretty well known by everybody here and certainly one of the most well known pharmacologists that we interact with and known throughout the world. So Howard, thank you for leading this.

[18:20] remind me to record that and send it to my mother. So thank you very much. It's been a fantastic day. Looking forward to the rest of today and of course the sessions tomorrow. But one thing I love about this conference is it's not one of those here's the

[18:40] latest thing I'm doing type of conferences, it's where's the field and what are we trying to do as a group? And this session continues that theme. We've got some leading lights in the field that are covering a number of different segments of how do we adopt this stuff clinically. And we'll be hitting on some themes that have been consistent.

[19:00] from the very first talk and there will be some new ones that come out as we dig into these sessions.

[19:20] director of clarified precision medicine on the clinical front. Also board member of the personalized medicine coalition, full disclosure. So I'm biased positively towards this group. So here weighing one, no. Next up is so Tom Brown's well known to this

[19:40] audience. Chief medical officer at SIAPs since October of 2019. He's a faculty member at the Institute for Systems Biology that's based out of Seattle. CEO and president of the fairly new, I think it was last month or this month maybe even, real world evidence alliance. So that's exciting. We're forward to hearing more about that.

[20:00] that. And then was when I first met you, well I guess one of the times I first met you was when you were the previous director of the Swedish Cancer Institute in Seattle there. He's also a professor of medicine with all his activities. So in his spare time, no. Next up is Cassie Hyac. So Cassie has two

[20:20] Not one, two medical director roles. So he's making the rest of us look bad. She's the medical director at Population Health Company Helix and also has a medical director role at Sanford Imogenetics. So within an integrated health system.

[20:40] as well as with a population health company. She's an internal medicine physician and been very active in this field at the clinician level, at the general health system level, and then at the population health level. So looking forward to those aspects. And then at the very end here is Cristina Aquilante. Cristina is a professor.

[21:00] professor at the University of Colorado. She's one of the directors of the Colorado Center for Personalized Medicine. She co-leads the system-wide pre-emptive clinical pharmacogenomics implementation across the university's health system and has been a major force for thinking about how things happen at academic centers and

[21:20] the rural setting or in less represented populations. And so this idea of making it so that all benefit as opposed to just people who happen to show up at University of has been a key theme to the last X number of years, won't say how many, not too many, of Christianist careers. So really excited.

[21:40] about that. I want to thank you all for coming. Now the reason I did the introductions there is I didn't want them to get stuck on the stuff that you could read about, but really get to, so you can get to know them as they're thinking about precision medicine. So Tom, we'll go ahead and kick things off with you. I wonder if you could talk about your clinical adoption focus and how you're thinking about

[22:00] adoption in this field.

[22:20] precision oncology context. And on the patient care side, one way to look at this is, I'm a lifelong developmental therapist in the sense of primarily working in the realm of classical clinical trials.

[22:40] early phase clinical trials, in particular, phase 1 work. Classical clinical trials are terrific and they're essential, but they represent a very narrow spectrum of the broad population. And one of the advantages of real world data is one sees it's a more inclusive view.

[23:00] and to the broad population that we serve, whether one is talking about race, ethnicity, age, socioeconomic status. So on the patient care side, we leverage real-world data to put a sophisticated mirror in front of clinical practices. You know, if you ask any clinician, and I'm sure Kathy and Kristina will

[23:20] would agree with this comment, all of us, if you ask us, how do we do, for example, when it comes to BRCA testing in women with triple negative breast cancer? And everyone feels they're at the top of the class and that it's done routinely. The truth is that when one looks at real world data, that is not the case.

[23:40] And there's an exercise where you can put that mirror in front of one's practice, identify gaps, and then understand those gaps, develop interventions, and then monitor improvement. Very much like the approach that Gary and Arushi just presented to us. On the other hand, on the regulatory side,

[24:00] side and again this is to me a very exciting realm. I mean the truth is that the FDA has in oncology certainly has done a wonderful job in terms of accelerated mechanisms of approval. But with that has come a shorter period of time prior to approval.

[24:20] A fact that fewer randomized clinical trials are utilized with new drug applications. So it's an opening to leverage real world data, either in a supplemental way or in some cases even as part of confirmatory evidence. And then with that rapid approval, there's much to still be discerned post-approval, whether

[24:40] in the realm of safety or efficacy, the latter related to label expansion. So all of this is relevant to precision medicine, precision oncology, personalized medicine, because certainly in cancer the majority of new agents are agent-

[25:00] with an associated companion diagnostic.

[25:20] genomics broadly in the use of patient care germline genomics. And really how we're thinking about it is sort of an access, I think, perspective in that as Tom just pointed out, many patients who actually have indications for germline genetic testing for specific indications.

[25:40] Most, in fact, don't actually get the testing. And there are a lot of barriers that exist that contribute to that. The cost, clinician understanding of what the result may bring, difficulty even ordering the test, support in, you know, returning results to patients. And so the way

[26:00] We set up our programs are large scale research studies with the clinical return of results that facilitate sequencing of the population for a given health system. And that data is digitized for reuse over time for specific clinical indications. So right there you're

[26:20] eliminating a lot of the barriers that we see as really challenges to clinicians using this in patient care. But I think it's important to kind of piggyback on some comments earlier. It's not just about the test. It's about the program that you set up around the test.

[26:40] So, that pharmacogenetics example is just a beautiful one. You can't just say here's the CYP2C19 genotype, that's just not going to fly. You have to set up mechanisms to help the clinicians use that at the time that they need that information. So again, right drug, right time, right patient comes into play. So we're really

[27:00] thinking about this broadly and trying to ensure that all patients who need this in their care deserve it while making it readily available for the new applications that are arriving to us daily. Thank you very much. Christina, you've been leading a lot of the efforts for the

[27:20] last while in terms of making sure academic medicine doesn't get stuck in the lab or really gets out into the clinic. I would love to hear your thoughts about clinical adoption and where you're focusing. Sure, thanks Howard. So when we give a little bit of a disclosure, I was a late ad so you're going to get a bit of stream of consciousness here. I called you this morning. It was this morning.

[27:40] Just bear with me, everyone. Someone from Mayo Clinic dropped out late last night. And Christina, unfortunately, had her name on the attendee list. And when I got to her, I knew exactly who I wanted on the panel. So she gets big credit for that. Yeah. Sure. Yeah. So I'm really thankful. Thank you for asking.

[28:00] asking me. I'm really excited to be here to be talking with you all about adoption. You know I spend, I've spent most of my waking hours over the last seven years thinking about how to implement, and I'm going to be mainly talking about germline pharmacogenomics into clinical practice and how to get clinicians to use the information to positive.

[28:20] effectively impact patient care. And from the academic medical center perspective and also a large health system that cuts across a state that serves a lot of different types of patients, you know, how do we do this most effectively? And some of the guiding principles that I can share with you, there's really been three main.

[28:40] main ones. We want to do preemptive testing. So how do we test patients before pharmacogenetic tests? How do we do the tests before that information may ever be needed? And second, how do we integrate that into the electronic health record in a way that it's maximally useful over the patient's journey?

[29:00] at the healthcare system. That's been really critical. And then lastly, how do we make this information as easy as possible for clinicians to use? And all of these themes you heard already today, and I can say being boots on the ground, these are all real. This is all happening.

[29:20] The way that we've decided to approach the implementation at our institutions is really kind of two different strategies. Our first strategy is we take, we focus, we call it a more targeted strategy, we focus on high risk patients, so patients who are at high

[29:40] risk of adverse drug events due to pharmacogenetic variants. And we started a pilot in patients who have GI cancers and we conduct preemptive pharmacogenetic testing and we implemented clinical standard of care. So it's not a research study. It's a real

[30:00] world pilot. So I was really happy that we're doing what Dr. Sherman talked about earlier. And the key here for us is putting this information seamlessly into the EHR so that clinicians can use it easily. The clinical tools are at their disposal.

[30:20] And it's really broken down barriers. We know that there are, we know that there are gene drug, gene chemo associations in terms of adverse drug events, some of which can be life-threatening. And we know that clinicians, oncologists are not, at least at our institution, we're not testing

[30:40] those patients routinely. And once we broke down those barriers and we made the test available, we put it into their workflows, we coupled that with clinical decision tools, they've just gobbled it up and they are just asking for more. And so that's really exciting for us. The other end of the spectrum is more populated.

[31:00] screening. So how do we improve access across a variety of different patients? And the credit here really goes to Kathleen Barnes, who's the founding director of the Colorado Center for Personalized Medicine and our Biobank initiative. So patients at our health system can choose to participate in a Biobank.

[31:20] And as part of that biobank, we return clinical pharmacogenetic results back into the electronic health record. So right now we have over almost 250,000 patients in our health system consented. We've returned results to over, I think, 36,000 participants.

[31:40] I think that amounts to over 220,000 individual results returned. And the clinicians are using it. And it's because we thought about it as a whole. Again, the buzzword that's come up already today, holistic, that holistic approach. So both a targeted and a broad strategy.

[32:00] strategy to try to get at some of the higher risk patients and then more of that population screening approach. Yeah, that's great. Thank you. So Dr. Heineken, maybe I'll start with you for this next question. We've heard all three of you have alluded to some of the endpoints that you think matter. And I wonder if you could talk a little bit about

[32:20] What are some of the really going to move the needle? We all have some endpoints that are easy to grab and make us feel good. But what are some of the endpoints that you're really focusing on that's going to make the kind of changes that everyone is going to care about? Yeah, I think a lot of it is consistent with what's been said.

[32:40] said today, it's really sort of making it easier for clinicians to use and apply in their patients. I think about, you know, when we work with health systems, the question of whether or not they're going to incorporate genomics as part of their population, health, and even broader strategies is

[33:00] not an if, it's a when. And so more and more of these systems are readily employing these approaches. When we get into the system, the clinicians are even very excited about it and want to use this information. But there is a lot of fear because for the overwhelming majority

[33:20] of the clinicians in practice today, they don't have experience or exposure to this type of information in the care of their patients. So it really just goes back to, I think, how do we create a little bit of an easy button for making this available. And so that

[33:40] goes anywhere from determining what is the appropriate informed consent for a test. One thing we run up against with certain testing, pharmacogenetics is one of them is in some states they're pretty significant restrictions on how you consent patients for this. And for a clinician who sees 20 patients a day and

[34:00] you want to do more, you want to be able to offer pharmacogenetic testing for, you know, for SSRI therapy. For example, it is very big ask to have them consent to patient in the way that we are requiring today in order to get that testing.

[34:20] feel that that information could be helpful and useful. So how can we work with the restriction in place today? How do we work with tools in the AHR to really help facilitate ordering that test? I think as you pointed out, the backend decision support of that is also really going to move the needle.

[34:40] Just again, returning a result is really just not enough. We have to be able to ensure that clinicians have a support net to actually apply this information.

[35:00] That's a large ask for any one health system to create the content and decision support required for that. So I think one of the things we're thinking about in that way is how do you create a network of systems where that content can be more readily shared.

[35:20] and updated in an evidence-based manner. Thank you. Dr. Brown, I know you've thought a lot about these endpoints, thinking back to your time at Swedish, your time now working across a number of different systems with your real world evidence, either with SAHPS or the Alliance. Love to get your thoughts on.

[35:40] on what are some of the endpoints that really we should be, maybe are undervalued now, it should be focused on more. I think in listening to Kathy and also with Christina's focus, we've heard throughout the day when it comes to pharmacogenetics, pharmacogenomics, that there's a real disequilibrium.

[36:00] equilibrium between what's known in terms of information that's immediately applicable and the actual application. It seems as though, I just wanted to comment at first, that that disequilibrium is probably due to the economic drivers that have also been.

[36:20] alluded to. And it dawns on me that one of the main steps that we should consider, and certainly that the PMC should be promoting, is again that holistic view that includes concepts of financial toxicity and as an actual

[36:40] endpoint in terms of framing the complete picture. I did want to answer your question. I think that certainly in the real world data realm, we spend a lot of time focused on what are the proper outcome endpoints.

[37:00] that would be equivalent to disease-free survivorship or progressing-free survivorship. Those are hard to discern when one is looking at real-world data. So we use endpoints such as time to next treatment, time to treatment discontinuation. Of course, survival is always the ultimate input in many.

[37:20] endpoint in many ways. But I think it needs to be recognized just as in classical clinical trials that the endpoints will vary depending on the exact hypothesis-driven question that's being asked. The other comment I want to make is that we focus

[37:40] a lot on surrogates for precision medicine, personalized medicine such as number of patients tested, and number of times that targeted therapy is used with an associated companion diagnostic. And one thing that certainly through our work,

[38:00] discovered is the importance of paying close attention to an accurate assessment of who has not been tested and why, and then developing strategies to address that. It's one thing to focus on the results of tests that have been done and is the appropriate therapy being instituted.

[38:20] And increasingly, in what line of therapy is that therapy being instituted? Which the whole issue of sequencing and compimatorial approaches is a great challenge for personalized medicine. So I mean, there are other things to talk about, but I'll stop there. Yeah, no, thank you for that.

[38:40] You, Christina, your center serves the university, it serves your health system, and it serves your state. As you think across those different, and maybe more, as you think across those, what are some of the endpoints that might broaden our mind and the audience's thoughts on?

[39:00] on how we should be looking at success in this adoption realm. Yeah, it's tough. I mean, there's not a perfect answer. There's not a perfect set of endpoints that will work everywhere. When we talk about pharmacogenomics, the thing that we're talking about is,

[39:20] I'll just give you our experience with what the C-suite asks me. They want to know, was the right thing done at the end of the day for the patient? So those are some of the metrics that we're trying to capture. We tend to look at that. Our tools are used

[39:40] across all of our hospitals. And so we look at how those clinical decision support tools are being used, by whom. We look at the clinical actions that are being taken, what the effectiveness rate of our tools is.

[40:00] We're also looking at trying to talk to our clinicians to see how they're doing with the different implementation pieces and We haven't done such a great job of this, but we're working on it But trying to get feedback from our patients as well over the long term

[40:20] I think one of the interesting things, this is kind of the flip side, but when we started, and I'll talk about the biobank, we're going to release results that the patient's clinician did not order, right? The clinician did not order those tests. We're just going to put these back into the EHR. The patient might not

[40:40] remember they ever participated in the bio bank. The concern was oh my gosh you're gonna have millions of phone calls and emails and everything and can you guess how many how many questions we've received so far? The way you're teeing it up it has to be a low number. It is a super low number. Right now it's less than four

[41:00] authority from clinicians and providers that have kind of dialed into our hotline or typed in. So I think there's this conception that people are going to be really afraid of using the information. And that's not true if you set the system up right. If you build out this holistic

[41:20] approach, if you build out these tools in a way that works with how the clinicians work and in a way that the clinicians can understand without a ton of pharmacogenetics knowledge, right? And so it's been really fascinating to see that. And they want more.

[41:40] And they'll call us and say, is this applicable to these drugs? And you have to be like, no. Or we had an email today about an additional gene that people were interested in. So the appetite is there. I think it's just the delivery, how you get it to them.

[42:00] Yeah. And I think one of the things that we've seen both on the academic side as well as the work we've done with clarified precision medicine has been the who's got your back component. Yeah. It's rare for one of our clinical colleagues to want to even know the gene name. Yeah. They all sound like Star Wars characters anyway. They sure do.

[42:20] So I don't really care if it's CYP2D6. I want to know what to do. I want to know what to do. And that shift in terms of clinical adoption really brings it back to the practicalities. And I think that's one of the useful pieces. And then them knowing that someone has their back.

[42:40] Often they won't necessarily even, like you said, they won't necessarily even use the mechanism, just knowing that someone is there and can help because so few people have been trained in this area. But when it talks to, you know, do you want pharmacogenomics? No. Do you want better patient safety and save yourself some time on drug select.

[43:00] Yes, you know, so it's some of those aspects. Tom, you were going to add something? I was going to comment in line with what you're both saying. We have a range of health system partners and it's interesting to note some have rather centralized precision medicine, personalized medicine programs in terms of the very details that you're speaking.

[43:20] speaking of. And I think we would all agree that they have tended to be at the top of the class in terms of efficiency and implementation of personalized medicine. And, you know, centralization for many of us physicians is a bad phrase, but I think

[43:40] I think in this case it has helped tremendously. I mean there's a balance to it in terms of providing sort of a supportive framework of the type that you're talking about. There are probably more personalized medicine experts at this meeting than there are not at this meeting. So please be very careful as you drive and travel home.

[44:00] We need you. We need you badly. So one more question before I open things up to the, you know, any audience questions. And Dr. Brown, maybe we'll start with you on this one. And that's really kind of a magic wand question. As you're thinking about the different obstacles, and one thing I really like about the way this meeting has played out so far

[44:20] far is the point has been made that there's not an obstacle. There are multiple different ones, everything from knowing to do something in the first place all the way through to how to pay for it and a bunch of things in between. But as you're thinking about solving this with a magic wand, what is the

[44:40] ones that you think would be on the top of your list. And I'm not even talking about which ones would cause faster adoption. Which ones would cause more impactful adoption? And I'd love to get your thoughts to start off. Yes, a medical oncologist. I'm not a pathologist. My colleague, Dr. Anna Berry, who is a fine molecular pathologist, is here today. And we all

[45:00] talk about this point because you know there's an obvious solution that in certain settings is a well-placed solution and that is reflex testing right to consider as part of the initial pathologic interpretation the biomarker profile and one can certainly argue

[45:20] that we have to be careful in terms of where reflex testing is implemented and where exactly is it done. But I think for many busy clinicians that would be an important step to have that being a more universal thing.

[45:40] the other item, again, whether you call it a molecular tumor board or a disease-specific site tumor board or general oncologic tumor board that has an element of this kind of molecular profiling coordination, both in terms of when study should

[46:00] should be ordered and then when tests should be ordered and then what should be done about the test results, who should be tested and who need not be tested. The other comment I want to slip in is, as happens in developmental therapeutics, we move upstream to earlier stage disease, to new adjutant settings and of course,

[46:20] ultimately even prevention or early diagnosis settings. And that too adds a layer of complexity because it's not only what testing for what disease, but when that testing should be done. And again, when one has either reflex testing or a molecular tumor board functionality.

[46:40] in some way those things can be facilitated, those decisions can be facilitated. Cassie, as I pass you this Ritz-Carlton number two, I mean magic wand, what are you thinking of in terms of what are some things that you'd love to just weren't there in terms of going forward?

[47:00] forward. I'm mentioning that not because that will happen. It is a pencil, not a one. But maybe there's somebody in the audience who has a clever way of really tackling this in ways that four of us haven't thought about. Yeah. I think, so I'm an internist at heart and prevention is really my passion. And when I think about it, it's really hard to think about it. And I think that's a really important thing.

[47:20] think about how genomics can inform that. We have always emphasized, I think, identifying those at highest risk, but how do we really leverage the entire spectrum of the population and identify patients who will follow the lower risk category and, I think, about the time that will come.

[47:40] I think, where we can start to maybe delay screening or change to a lesser intense screening approach for patients whose genomic profile may indicate they don't necessarily need that. And I think from the prevention standpoint that's exciting.

[48:00] I think from the sort of the lift for the healthcare system, when you start to really stratify your solutions to the patients who need them the most, you improve access for all. And so I think it'll come. I don't know that we need a magic wand for it. It's just time.

[48:20] back down. Dr. O'Quality, what's your magic one moment that you like to...payment. Right now? Right? Yes. Yeah. So you know I think you know evidence is there for a lot of personalized medicine, a lot of pharmacogenomics, but but payment is not.

[48:40] That's just the reality right now and it's extremely challenging and Some of us are lucky to work at medical centers and health systems that are putting in the dollars to make this happen because they believe it's the right thing to do and They're trying to move the needle on on their own But we know that you know if these tests were paid

[49:00] for, reimbursed, and more people could have access to them. That could substantially reduce healthcare costs and deliver better care. And when you think about it, the cost of adverse drug reactions and ineffective therapies, it's a couple of billion dollars a year in the U.S.

[49:20] And I just think about those trade-offs and we could be kind of chipping away at that. Not that pharmacogenetics will solve all of that, but it's another clinical tool that we have at our disposal to prevent some of these bad reactions. And I'll give the example of the

[49:40] GI oncology clinical pilot that we started. We focused on this high risk population. Our GI oncologists are amazing, great collaborators. They're like this is great, thank you for doing this. The first patient we tested had a polymorphic genotype and they were going to be put on

[50:00] a T-can which is a pretty toxic chemotherapy. And that combination they could have potentially had a life-threatening adverse drug reaction. And from our pilot, that first patient, the clinical decision support we had in place, the dose was reduced, the patient did just fine.

[50:20] no problems. And I get goosebumps when I think about it because just that one patient, that first patient in our pilot had a positive benefit from that test. And I know there's a lot of talk about cancer at this meeting so far. And cancer has impacted probably all of us in some way.

[50:40] And when I think about it, if we can use this information to make a patient's journey through their cancer just a little bit more tolerable, it just seems like a no-brainer to me. The other, can I say one more magic wand? Please. Okay. It's also- An unlimited number of wands.

[51:00] Unlimited ones, yes. It's to, if I could erase some of the exceptionalism that we have about personalized medicine and pharmacogenetics. Not to keep treating it like it's something super special. It's really not, right? It's just another piece.

[51:20] of clinical information that can help improve clinical care. And I think sometimes we need to get out of our own way and just think of it as another lab result, another test that we can use to positively impact medical decision making. And so I think that that would be my second magic one. Thank you. Tom, you want to add something?

[51:40] Well, Christina, first of all, thank you for those comments. And you're triggering the thought that I think PMC could be very helpful with, and that is advancing the way we think about illness. Because I believe in the not too distant future, as Kathy was alluding to, that we'll look at illness as

[52:00] being the risk of illness. And that has genomic input. It also has environmental input and socioeconomic factors and so forth, as we're all aware. But I think that shift in mindset will probably promote a shift in payer policy, for example. So we have time for a couple of questions.

[52:20] questions and then we'll close it off with a few last comment from each. But if anyone wants to come up to the microphone or the microphone can go to them, please do. They don't have to be difficult questions. They can. It looks like Dr. Pritchard might have one to start us off.

[52:40] And thank you to all of you. At Samford Health and at the University of Colorado and then even more importantly at SIAPS, which deals with lots of different providers, you guys have the benefit of understanding the implementation challenges and doing something about it. And if you could wave your magic wand, you would.

[53:00] that you are probably working to do those things. It's not going to happen overnight, but you are working to do those things. My question is, what can we do for those systems that doesn't have leadership like we had represented up on the stage in this conference room? How can we scale some of these strategies? The use of molecular tumor

[53:20] board for more regular reflex testing for pharmacogenomics. How do we scale that to the whole health system? Cassie? Yeah, I can start. Well, so in our work at Helix, that's actually kind of the emphasis of our work, is really bringing this to systems that maybe don't even

[53:40] even have clinicians experienced in genetics, or maybe it's a small team. And so how do we design this in such a way that it's applicable to a broad set of health systems? And so the way we're doing that is taking the experiences that we have

[54:00] from across the different health systems in creating best practices, creating a network of those systems to share those best practices across the board, not only hopefully among each other, but even more broadly. So I think having PMC as.

[54:20] as a place to sound off some of those best practices will be pretty beneficial to the broader implementation of these programs. Thank you. Ray Lawrence from CASEL. Hey, so you mentioned kind of the exceptionalism that we have with pharmacogenomics potentially.

[54:40] being held to a different evidentiary standard, different level of evidence, that kind of thing. So I guess I have two kind of related questions. The first is really, how do we overcome that? Right? And there's not an easy answer. I get that. Right? But also number two, you know, when we do have successes with a specific health system or with a specific payer, how

[55:00] can we parlay that into success with other places who also want their own data to be done on this same set of patients? I can start with the first one. I think for the first one, we just try to keep it simple. We just want to practice evidence-based pharmacogenomic medicine.

[55:20] and Kathleen Smitling, she's heard me say this like 100 times. And just looking at the data that's available, guidelines that are available, what's in the FDA label. We don't try to do, we personally don't try to do fancy combinations or anything like that. We keep it simple, we focus on what the evidence is strongest for.

[55:40] We deliver it in the simplest way to our clinicians and we just try to keep it simple. And that's, I think, worked pretty well for us. I'll let folks. I think when it comes to reimbursement issues, I think many of us have found some

[56:00] success in leveraging outcomes from molecular tumor boards or tumor boards with pairs, and this would apply equally to pharmacogenetic questions for sure. I think in general, when it comes to and it applies to the question that Kathy answered as well.

[56:20] Physicians, certainly, and by extension, healthcare administrators are generally evidence-driven. So we've very much focused on work such as what Arushi and Gary presented to, maybe in a small way, address a relatively simple question and show the benefits.

[56:40] of not only being aware of the challenge, but showing the impact of certain interventions. So I think that it's all about evidence-based approaches. Thank you. I think a lot of it also is coming down to...

[57:00] showing examples that aren't exceptional. So we have examples where the mechanistic pathway for a drug interaction and a pharmacogenomic interaction is exactly the same, yet one is blindly accepted and one has a lot of fighting. And it's like literally the same mechanism. So part of it is just...

[57:20] taking the the the Hippocratic oath rather than the Hippocratic oath in terms of so we have two minutes and four seconds left I wanted to get each of you to channel your inner ed and talk about very briefly the positives for

[57:40] clinical adoption because you want to stand positives and there's plenty to be positive about. So maybe Christian, we'll start with you and Morquh Ray Dantaton. I think the time right now is really exciting and it's been my experience has been extremely positive and it's been

[58:00] And the kind of the saying, if you build it, they will come. And that is what has happened at where I'm at. And it's just been really great to see people want this information and use it and see it documented in their notes and patients positively impacted from therapy. And I think that goes.

[58:20] goes all to kind of the thoughtful design about the patients you're targeting and that kind of approach of how you're going to deliver that information to clinicians and to patients as well. So I'm not usually an optimist, but I'm very optimistic about what's happening right now in terms of adoption and implementation.

[58:40] Yeah, I would echo that optimism. It seems like in the last couple of years the tone of the health systems that we approach has really shifted from, like I said before, it's not an if, it's a when of incorporating genomics more broadly in patient care and so the enthusiasm and the willingness to

[59:00] to start to think about genomics as part of longer-term strategies, that things that don't happen overnight is palpable more now than it has been. I think also the interest from clinicians in this space, and I'm talking from primary care

[59:20] to subspecialty, I can think about a time when I was at St. Brett Health and we launched our program when our program was pretty broad and I'd have some subspecialists say, well, you know, there's no application for genomics in my field. And my response was always like, try me. I'll come up with something and I think now you don't hear that anymore.

[59:40] So I'm very optimistic.

[01:00:00] certain way, it was always 5FU in some fashion, everyone with lung cancer another way. And now we've evolved and we see it in our day-to-day practice where we think of the molecular fingerprint of one's tumor that one's colon cancer might be in some ways similar to someone else's lung cancer, a little bit of a problem.

[01:00:20] of hyperbole there. But the point is that we're now thinking of the uniqueness of the molecular profile of the patient's tumor and the patient themselves, which of course gets to the genomic aspects. And I think that trajectory defines the progress that we made. Well, please join me in thanking our panel for this session. Thank you.

[01:00:40] It's now my pleasure to ask PM

[01:01:00] MC board member Liz O'Day to come to the stage to introduce our next speaker. Liz is the CEO of Alaris, a diagnostic company in Waltham, Massachusetts. Liz.

[01:01:20] Ladies and gentlemen, esteemed colleagues and distinguished guests, it is both an honor and a pleasure to stand before you today to introduce a trailblazer in the field of personal

[01:01:40] medicine. Dr. Amy Abernathy couldn't be a better speaker to close out this first day of what has been a truly inspiring conference which also happens to commemorate 20 years of personalized medicine. Amy is a visionary leader whose contributions

[01:02:00] have not only shaped the landscape of personalized medicine, but she has also ignited a fervor of innovation and transformative change. Dr. Abernathy's journey is nothing short of remarkable. With a background that seamlessly combines clinical expertise and groundbreaking research, she has emerged

[01:02:20] as a driving force behind the paradigm shift towards patient-centric and data-driven healthcare. Her relentless pursuit of excellence has not only earned her widespread acclaim, but she has significantly impacted the lives of countless individuals across the globe. She is currently a member of the American Health Organization, and she is a member of the American Health Organization.

[01:02:40] Currently the president of product development and the chief medical officer at Verily, where she leads teams in the development and delivery of products that bridge the gap between clinical research and care. Previously, she was the principal deputy commissioner for the FDA and the agency's chief information officer. Prior to her role

[01:03:00] Dr. Abernathy was CMO, CSO, and senior vice president of oncology at Flatiron. And previous to that, she was a professor of medicine at Duke. Dr. Abernathy is a hematologist, oncologist, and palliative medicine physician who has authored more than 500 pom-

[01:03:20] publications. She holds a B.A. and biochemistry from the University of Pennsylvania, an M.D. from Duke, and a Ph.D. in evidence-based medicine and informatics from Flinders University in Australia. Amy is renowned for her business acumen and scientific prowess and is someone I truly

[01:03:40] admire. Of her many, many achievements, I'm sure that there's one that's on the top of her list or at least on Ed's list and that's Amy was is a former board member of the PMC. Amy continues to support our mission through her work, through her example, and through her friendship.

[01:04:00] And so without further ado, please join me in welcoming Dr. Amy Aperthy to the stage.

[01:04:20] 20 years in at PMC. Wow, right? And as we think about where we are in personalized medicine and the ability to achieve precision health, it's made possible by your dreams, it's made possible by your innovations, and we're seeing it show up over and over every day.

[01:04:40] right now. And just looking at the number of innovations coming out right this second is spectacular. Just think about it. The UK just approved a CRISPR-based therapy. We think it's going to be approved in the US very soon. Wow. About a year ago, Matt Herper wrote a piece in

[01:05:00] stat news called Biology Century and really it highlighted the challenge we were going to run into trying to achieve the true potential of biology century. And what Matt highlighted was that the choke point is going to be

[01:05:20] evidence generation. When I say evidence generation, I mean clinical trials, I mean real world evidence, I mean the totality of the data and the evidence that helps us understand does this treatment work, is it adequately safe and effective, for whom, when should I use it, and it is the information that we need

[01:05:40] as regulators, but also the information that payers need, clinicians need, patients need in order to figure out what to do. Interestingly, as I look at the landscape of evidence generation, it also is an important reminder that the landscape of evidence generation itself is changing.

[01:06:00] I was just listening to Tom speak and remind us about well what that's been going on in rural evidence and rural data, but as we think about where we're going into the future, we're going to continue to need to change how we think about evidence generation.

[01:06:20] In cell and gene therapies, we understand the challenges where we see incredibly powerful interventions with huge effect sizes, but we're all kind of questioning this issue of off-target effects. What's going to happen across time? A conversation last week around lymphoma and CAR-T. So we're really

[01:06:40] Really in an interesting place, and in January of 2020, I was responsible for putting forward FDA's guidance of needing 15 years at least of longitudinal follow-up after cell and gene therapies. Just think about that. 15 years of follow-up for these new treatments.

[01:07:00] Once they've gotten over a line where that burden of follow-up lives on the shoulders of the people with unmet medical needs who needed the treatment in the first place as Well as the cost of the system and the cost to trying to continue innovation. We see the same challenges and artificial intelligence-based solutions

[01:07:20] How are they going to be regulated? What are we going to do? We know LLMs, they hallucinate. So the AI-based therapies can go wrong and often do across time. Evidence generation and its demand is coming from all corners as well.

[01:07:40] Medicare has certainly stepped up its focus on evidence generation. What we've seen, whether that's the IRA and the expectation of data for negotiations, or coverage with evidence development activity that, for example, showed up in the landscape of Adj, Kinema, this story,

[01:08:00] Medicare is also getting into the conversation. It was playing out, as many of you know, across the globe, for example, through health technology assessments, but now we're seeing this also in the United States. As we think about the role of novel diagnostics

[01:08:20] the context of evidence generation, we have this opportunity to leverage novel diagnostics, for example, for new endpoints. We've got the opportunity to leverage diagnostics to help monitor people across time and apply that into clinical care, but we've got to figure out how to do all of this. And I always find this interesting.

[01:08:40] This is actually a framework put forward for AI, ML, so artificial intelligence-based software devices, in 2019. In the bottom right-hand corner, the reminder here was not just the importance of getting clearance for software-based devices. We're not exactly sure how that also is going to happen quite yet.

[01:09:00] importance of performance monitoring of these software devices across time. Think about that. If AI-based products can hallucinate or get better and better and better until things go poorly, we got to keep check on that and we need systems that are going to allow us to do that. And then finally, very recently, got some

[01:09:20] Just a couple weeks ago, the Reagan-Udall Foundation did work on behalf of FDA to start to focus our attention more and more on the post-marketing space and what that should look like, recognizing that importantly our task is not finished once we hit the regulatory line. A drug or diabetes disease.

[01:09:40] diagnostic, being cleared or approved, we still have work to do to follow it as it gets into the market and how do we do that? Well, you've heard me talk about this conversation around the shifting evidence generation and landscape, but I submit to you this story is accelerating in front of our.

[01:10:00] What we have seen is the move from the figure on the left, a formulaic, phase 1 through 4 clinical trials, usually a regulatory approval after phase 3, to something that looks more like the figure in the middle, where oftentimes those regulatory approvals are coming with a smaller corporate

[01:10:20] of information. This might be shorter clinical trials or single arm trials or fewer numbers of patients, but the move, the shift to our understanding how these treatments perform into the post-marketing setting. There's another thing important about this figure. So this figure shows you that our need to understand medical products

[01:10:40] intervention to improve personalized health is longitudinal. It has to happen across time and that we need to keep check across time. This figure also reminds us that the demand for evidence is going up not down. So we're going to need to come up with efficient systems to make this happen.

[01:11:00] And further, we need to make sure that we have information that represents all people. Personalized means for me. And if we have evidence that doesn't reflect for me, we've missed the mark. So I submit to you that a critical way of

[01:11:20] getting there is to connect the clinical trial space and the care delivery space and the interstitium, the parts of our lives that exist in between, so that we can use all data to create longitudinal data stories that in fact represent individual personal journeys.

[01:11:40] and where longitudinal data stories then become the foundation on top of which evidence generation can happen. Now in this figure, the clinical trial in this particular figure is in purple. This would be a study with formalized protocol, usually at a time of consent.

[01:12:00] with data that's come from other settings, such as the electronic health record, that's in green. It's typically kind of episodic. And practically speaking, episodic data isn't necessarily captured at specified times and may have different features than that which is specified in the clinical trial. And then other information that

[01:12:20] comes in the interstitium, the information in the yellow, might come, for example, from sensors in our watches, or comes from environmental data, or perhaps large genomic data sets. And these data sets have the opportunity to now be included in this longitudinal story. But if we're gonna do this, we have a lot of time.

[01:12:40] masks at hand. And I submit to you that critical to pulling this off is figuring out how we connect together, the research systems and the care delivery systems, and do so in service of the patient to get us going forward. If we can do that, if we can develop better mechanisms

[01:13:00] to leverage all available data, it's going to make our clinical trials processes better. We'll be able to more seamlessly, for example, launch clinical trials within healthcare delivery systems in ways that we're just starting to see happen in the United States right now, but have great examples across the world, like the recovery trial in the UK. We'll be able to more

[01:13:20] easily recruit patients when we need detailed understanding, such as a formalized clinical trial, and then go back to usual care in much more efficient ways. We'll be able to coordinate the clinical trial activities with care activities so that you're not doing two different things for the same person. What about we can

[01:13:40] use some of the regularly collected data in the context of care to fill in the clinical trial data set and then just add in the new data elements that are needed. And certainly we'll be able to have more real-time insight in what's happening in terms of the efficiency of evidence generation and how to improve on that. I submit to you that if we really start getting

[01:14:00] getting this right, we have huge efficiency and practicality gains in conducting evidence generation and conducting our clinical trials. So if we're going to do this, if we're going to connect data, we're going to have to pay really close attention in the end.

[01:14:20] September 2021, FDA put out guidance about real-world data and specifically about the issues that needed to be addressed. One of those elements in the guidance was around the importance of understanding data quality, issues like missingness, data conflict challenges, when data might be

[01:14:40] erroneous and being able to understand it, improve on it, and document what's going on. Another thing that the FDA said during that period of time is it's important to figure out how to link together data sets in order to make sure that we can fill in data gaps. But if we're going to do that, if we link together

[01:15:00] other datasets, oftentimes we end up with conflict coming from the datasets. I always like to use the example from when I was at Duke where we had 11 source systems for gender. This was back in 2007, 2008. For all 11 systems said Amy was female, Amy's probably female. If seven said Amy's female and four said

[01:15:20] Amy is male, we got a problem. And importantly, that shows up oftentimes in every single variable. So we got a lot of work to do there. Also, if we're going to use these data elements, for example, for high risk tasks like formalized clinical trials, then during those circumstances, we need to understand the controls that were put around.

[01:15:40] collecting the data, like good clinical practice and auditability. And then if we're going to use the data element for less formalized processes, such as a post-marketing study, maybe you don't need those same levels of control. So we got to have ways of documenting the controls in place, meeting audit expectations and toggling that as necessary.

[01:16:00] And then also as we think about how to build these systems, we need to make sure that they can accommodate smart study designs. So you might have longitudinal registries where you can embed randomization. So suddenly your longitudinal registry becomes a phase III, at least for a period of time, if you've got the right set of controls. When you think about

[01:16:20] complex designs and be able to plan for that. It requires careful attention, but in 2023, we can build the infrastructure to make these things happen. And you can actually even take it up a notch from there. It's not just clinical trials data, plus, for example, data from the electronic health record, but you can now lay

[01:16:40] are on many, many different unique data types, sensor data, retinal imaging data, imaging including MRI or PET CTs, and importantly also keeping up with the context of what the patient has given permission to. So one of the interesting

[01:17:00] things about this story, and I'm going to show you in a minute what it looks like underneath the hood, is that informed consent, we currently think about it as a process. It's a static process that happens at a moment in time. There's a big document. It's pretty exhausting. It's important that we do this, and contractually this is how we get things going.

[01:17:20] But actually, that moment of interaction is a very special moment. And if we actually rethink about the consent process and use it as a handshake point for transparency and to walk forward, we can now think about using that interaction moment to now start the journey of how we do data collection differently in the future.

[01:17:40] And so the many things that we can do layering onto a system once we've started to think about how to build it. Now many times I find the conversation starts to drift into this kind of story of maybe everything can be decentralized clinical trials and we don't need to interact with the health systems at all.

[01:18:00] or decentralized evidence generation, and there are some really great circumstances where that's the case. But in fact, most of the time, when we're talking about high-risk health decisions, partnership of the overall system with providers, because patients need to partner with providers, is a very critical task.

[01:18:20] So as we think about building the infrastructure and longitudinal evidence generation, doing so with providers as partners is a very important part of bridging between research and care. I mentioned consent and you're probably going to find me talking about consent in the

[01:18:40] next two to three years way more than you probably wanted to hear about consent, but I actually think this is this fundamental opportunity that we've got to take things forward. It's the opportunity to develop a model of engagement. It's the opportunity to go from the 15 to 40 page form to something that's understandable

[01:19:00] actually continues to evolve over time. It helps people know how you intend to use data collected now and also to communicate what's going to happen with data and change that across time. It's the opportunity to get permission to recontact if there's an opportunity for you.

[01:19:20] to build trust and engagement. And if we think very differently about consent, it's suddenly the opportunity for the beginning of data stewardship, data permissioning, and walking with a person across time. Similarly, if we're gonna start to connect research and care, we need to get better outcomes. Like the fact that our clinical trials, we use our consent as a tool for

[01:19:40] use outcomes that don't always make that much sense in the context of clinical care is a real challenge. So we've got work to do as a community to build the kind of outcomes and endpoints that matter, but that also makes sense in the context of our practice because we've got to bring these two together.

[01:20:00] One of the things that we've been working on at Verily and I just mentioned is, for example, leveraging a medical grade watch, which we call study watch, where you have access to all of the sensor data as an opportunity to replicate the neuromotor exam in Parkinson's disease. And now suddenly you can have a patient learn to

[01:20:20] do maneuvers based on instructions from their watch and measure those maneuvers every day. And you don't end up with a neurologic exam once a month when they visit the neurologist, but you end up with a simulated neurologic exam every single day where you can watch the contours of the change in neurologic health across time.

[01:20:40] So, I thought I would give you the backdrop of how I've been thinking about this and then give you some examples of putting this in action. So practically speaking, it's very important to have infrastructure that forms that bridge.

[01:21:00] between research and care. Verily I think about this as infrastructure in the research workflow, infrastructure in the care delivery workflow, and infrastructure that supports longitudinal data generation. These are like nodes of activity through which then you have the

[01:21:20] opportunity to build longitudinal high quality data sets. As you build those longitudinal high quality data sets, it's very important to also think about the opportunity for that handshake point. The time when now it's prospective evidence generation as necessary, but connecting through tokenization and other

[01:21:40] capabilities to other data sets that already exist in the ecosystem. Once you start to build the platform, then you can do many things on top of it. Averily we think about this as building tools and those tools acting as nodes. Many different end-users

[01:22:00] purchase or use those tools, but in fact those nodes form infrastructure on top of which longitudinal datasets can form. So on the research delivery, and the research side of the equation, for example, site clinical trials management systems, the kind of system that large health systems and

[01:22:20] sites use to manage all of the trials that they're trying to keep up with and then using that as a platform through which you're able to do new evidence generation tasks. Tools in the care workflow. So for example, care management platforms for people with diabetes or cancer.

[01:22:40] patients and those platforms then also being able to be the mechanism through which you interact now directly with patients. And then the kinds of capabilities that you need for longitudinal evidence generation. Practically speaking, as a very specific example, we're

[01:23:00] partnering with one oncology. One oncology will use the clinical trials management system as that clinical trials management system for all 1,000 oncologists and their patients on trials. They'll be using one care delivery platform for the management of their cancer patients for cancer symptom management.

[01:23:20] management, for adverse effects, patient triage. And then practically speaking, they'll leverage, for example, other platforms for longitudinal evidence generation, and then we'll work together to start building registries, for example, lung cancer, colorectal cancer, etc. But underneath the hood

[01:23:40] The goal is that the infrastructure is flexible enough, it allows new things. So now suddenly what you're ready for is to build these longitudinal data sets in partnership with patients. The reason I wanted to mention informed consent yet again is because informed consent now also is the start

[01:24:00] It's the starting point for that handshake and continuously interacting with the person across time. It's also the starting point for data permissioning and data stewardship. So very practically as an example, electronic consenting should come with other things besides just DocuSign for trials.

[01:24:20] It actually should come with, for example, the readiness to do tokenizations as a technical capability to link data sets together. It should come with an understanding of what permissions did the patient give you, for example, to re-contact in the future or do other things. And it should come with the readiness to understand that our policy contacts.

[01:24:40] change like GDPR and the state of California and that those also need to be built into the system. What then can happen? You start building longitudinal data sets. Those longitudinal data sets now get layered in with newer and newer data types and you can start to do for example with a longitudinal lung cancer registry, the ability to do that is a big help.

[01:25:00] ability to, for example, do molecular subtyping or follow performance biomarkers or understand adverse effects of drugs. And then as the last concept, that really takes you now to a platform that allows you to do new things. So if you have longitudinal data sets, such as a longitudinal register

[01:25:20] that's been formed by bringing these different kinds of data sets together through a different infrastructure. That longitudinal lung cancer data set might be used for many different things by many different users in a syndicated format. But with the flexible infrastructure underneath the hood, if one user wants, for example, a

[01:25:40] derivative observational study with the right permissions for the patient and the right consent, you can add in new variables such as new patient reported outcome, our new diagnostic test and follow that person across time. So you've got a derived observational study. Remember, we're keeping up underneath the hood with what good clinical practice.

[01:26:00] and other research controls are in place. And so now in that same infrastructure, if you add in GCP controls, an exposure to investigational product, and the right consent, that infrastructure now becomes a phase one or phase two trial. And then again, that same infrastructure,

[01:26:20] Now with the addition of randomization, either to one arm or multiple arms as a platform, or a master protocol, now sets you up for randomized studies. And importantly, once the study is done for the formalized clinical trials aspects, for example, there's no more exposure to investigation.

[01:26:40] investigational drug, you go back to the baseline longitudinal registry study. And so that's where I think this story is going. The last thing I'll say to you is this is not a 2040 story. This is a 2024 plus story, but we've seen it play out in different ways around the world. I mentioned the recovery trial.

[01:27:00] Now we need the infrastructure that allows us to do it at scale, and I think you're going to see that starting to come together in the next year or two. So some closing thoughts. Really my question right now for myself is how to build for this story. How do we build for this transition to continuous evidence generation?

[01:27:20] across the life cycle of a medical product and build towards personalization. And also the importance of diagnostics in this story. Because diagnostics are going to be critical to understanding the details of individual people and how they respond. They're going to be important to understand how to monitor people across time.

[01:27:40] important for us to understand when something did not work, what else do we need to go and learn? And so it's going to be very important as we build that in. So with that, I appreciate everyone listening and thank you and I'll take any questions.

[01:28:00] I don't know. Any questions out there? Where did I stun everyone?

[01:28:20] give me one. Thank you for a great talk. I'm just kind of wondering what's it what's the business model for what you're talking about? It's a very interesting question. You know so as I highlighted the different products are nodes but every one of those products actually

[01:28:40] themselves has unique buyers. So for example, in the example that I just showed you, site clinical trials management systems are typically purchased by a health system, for example, or a large oncology practice. The care delivery model, so for example.

[01:29:00] what we would call light path oncology or light path diabetes, or you might have seen other care delivery models like Levongo out there and other things. They're usually bought by employers and payers to help care for populations of patients. What we usually see is life sciences companies and pharma tend to buy many of the care delivery products.

[01:29:20] I always call my E-consent product my switching station because shouldn't E-consent be the same thing as what we do for surgical procedural consent and HIPAA consent and isn't that really all the same kind of platform and that ends up being used or purchased by anybody in the ecosystem.

[01:29:40] that really needs access to electronic consenting or electronic patient-informed interaction capabilities. And so each of those has different buyers, but the importance is the readiness to now connect the datasets together. In terms of the longitudinal datasets, at least at Verily, we have three different kinds

[01:30:00] of buyers. There's some companies that come to us and say we need a longitudinal registry that looks like the following. The second version is companies that come to us like life sciences companies to say you know we we need to now start thinking about building a longitudinal data infrastructure.

[01:30:20] in oncology for longitudinal, for example, multiple myeloma, where we want to also be ready for derivative clinical trials. So we want to make sure we've set ourselves up to be able to do pragmatic trials in the future. And then where I think it goes, and that's just sort of like just give you some sense of where I think the world in this space is going to go.

[01:30:40] is that the other key buyer of that kind of infrastructure is ultimately going to be, for example, companies that need to do performance monitoring of AI-based products, datasets for not only developing products but validating products, etc. So we have a lot of work right now going on in the regulatory

[01:31:00] use cases there. Dr. Abenithi? Yes. Yeah. How do you see the funding of the incredible increase in diagnostics that are going to be required for all the post market? Is it the pharma companies that are going to fund that green blob you have on the right? That's a great question. Because I actually didn't realize.

[01:31:20] what a disparity there was between diagnostics and therapeutics until today. But it seems like a challenge.

[01:31:40] really critical and real here and we're already starting to see that. I also think we have a task at hand to figure out when we need the most sophisticated diagnostics and when we can use something that is less perhaps precise but a proxy that's less expensive in certain circumstances and be able to

[01:32:00] to understand how to calibrate between the two so that we understand the implications of data that comes from a scenario using less precise diagnostics and calibrate it across. And the calibration by the way is something that I've been actively working on because we're starting to see that become a need as we start to think about for example moving towards

[01:32:20] circulating tumor DNA and other activities is just an example. Thank you. What are the implications for economies of scale or network effects as you'd have more and more institutions using the same kind of platform like if half the hospitals in Minnesota were using a platform like you've shown on the slides or could you have a system develop or

[01:32:40] there are maybe incompatible platforms like Epic Concerners, so you'd have 1,000 hospitals on Verily and 1,000 hospitals on something from Optum. And then what about people that aren't part of institutions like they get their cancer care at Cedars-Sinai and their heart care at UCLA and their arthritis care at USC? Really important questions.

[01:33:00] So, you know, I'm working on how do we build it at Verily, but I'm actually also working on how do we make sure that we're building the national and international tooling that move us in a direction where essentially we have the right kinds of interoperability. So for us at least practically speaking,

[01:33:20] building the capabilities, but also making sure that we use common standards. So at least at Verily, we map everything to FHIR, and then we also are mapping everything to OMAP so that it actually is data standardization, and really that's where the ability to combine comes as opposed to, for example, it has to be that the software

[01:33:40] talks to each other. So that's one piece. The second piece I would say, at least from a Verily perspective, but not everybody, not every company does this, is being able to build interoperability through APIs and other software interfaces into what you do. And I've been very personally invested in how do we do, for example, smart on fire and

[01:34:00] How do we as a country push towards the kinds of interoperability activities that allow that to happen? That's the second thing I would say. Third thing I would say is that we have the opportunity to pick, and at the National Academy of Medicine I've been working on this idea of what I call central solves, things that we have to solve.

[01:34:20] and it will actually be able to unlock across different systems. And the two right now that I've been working on, one is really helping to update how we think about the informed consenting model, and the second is how do we document data quality so that we can do it in a similar way and then be able to look at different datasets with the same data quality document.

[01:34:40] documentation, etc., so that we can start to standardize that across systems. So those are some of the ways, but if you have other suggestions, I would love to know. Well, Amy, you've been bringing your vision and touring it with each successive PMC conference every year, and thank you for that.

[01:35:00] transparency and the vision of how we can get there and making information work. The one thing I haven't seen really evolve from this, you've talked about informed consent from the standpoint of uses of patient's data, but you've built and if you think about you know who's really good at delivering

[01:35:20] content back to consumers and patients. Google knows this world better than almost anybody and does it really well. And I don't necessarily see how evolving from this yet is mapped out, how the direct results from participation in a clinical trial, or even how good is my clinician providing to me this information that I need in my own care.

[01:35:40] care and decision making. And I think my opinion is that in the personalized world, health care world, it won't be achieved until persons have their own right to make those decisions on their own. So thanks. Here, here. Thank you. And I'll kind of give you two quick sort of thoughts.

[01:36:00] One is at Verily, because we had something called the baseline health study, project baseline, 2,500 individuals who've donated their life data story across time, truly blood, sweat, tears, microbiome, two-day in-person visits at Duke and Stanford. That gives us an incredible reference.

[01:36:20] data set that we can build against and it gives us sort of the ability to essentially validate as we're building infrastructure. But the other thing that it's done is that that group of incredible donors, volunteers, have also spent a lot of time helping us understand what does return of information.

[01:36:40] look like and how to do that in responsible ways that's understandable that tells the complete story. Because I think that your points about personalization really requiring the person to have their own information is fundamental. And we need ways of getting that information to people that's fully understandable.

[01:37:00] as complete as possible and where they can actually say no this part needs to be fixed as well and that that can also to connect to what to do. So that's certainly a part of the vision. I didn't talk about it here but you're a hundred percent right. Great talk. I think it's sort of really fitting.

[01:37:20] to end today's session talking about this handshake and sort of reframing and form consent to bring the patient back into the center stage, which is everything that we believe here at PMC. I'm wondering though if you can expand on how you think this will tackle certain health equity problems that we see and so that all these fancy

[01:37:40] tools we have can actually get to the masses and be more democratized. So health equity, I personally believe, has to be center stage in how you build and what you do. And so in January 2023, together with Vindell Washington, we opened a health equity center.

[01:38:00] excellence at Verily and the reason I did that was because I felt like what we needed was not only a verbal commitment but we actually needed a commitment to how do we build health equity by design into every product and what does that look like. My overall thinking here is that health equity is

[01:38:20] is a must if we're going to think about personalization. It's a part of that story. It also is about trust and trustworthiness. And so part of what we've been working on in the Health Equity Center of Excellence is starting on day one with

[01:38:40] essentially patient forums that are advisory committees that help inform product design all the way from the beginning. How do we work together with patient groups to actually tell us, fix this and do better here? How do we measure impact and change across time?

[01:39:00] and how do we think about what that measurement looks like? And then also how do we measure trust and trustworthiness? And we have about 70 user experience designers right now at Verily, and they're really very focused on this idea of building systems to measure trust and trustworthiness. And so in 2024, those measurement systems will be applied to all products.

[01:39:20] I'm curious how much you've had to invest in ontologies to make things comparable across different sites. That's even possible with patient consent. Interesting. Yeah, so it was interesting. When I came to Verily, informatics actually wasn't a

[01:39:40] piece of Verily and it just you know is kind of I think it was a byproduct of the fact that that's that messy part underneath the hood of health care that not everybody sort of realizes there until you start trying to work on it and and figuring out how do we think about investing in ontologies both internal

[01:40:00] early and also in the external community. So we make sure that everything is able to move along in essentially completely transparent and interoperable ways, which kind of goes back to your point. We've had to make some internal decisions. So for example, one big debate that was on

[01:40:20] going was do we use OMOP and if we're going to use OMOP how do we deal with certain use cases that really don't work within the context of OMOP. And then the other thing is that in doing this investment how do we invest in the internal education that helps to like move that conversation forward. And there's a long way to go but I feel like at least

[01:40:40] used in the last year, this has been sort of a pretty critical investment. Hi, question around your baseline study. I don't know if you can share any insights around anything you've learned about incentivizing patients to kind of pick up the stick and say, hey, I want to own this and move forward.

[01:41:00] everybody's Bonnie J. Adario, right? And we have kind of the general populace and so are there any insights around you know how to incentivize that, how to we've seen a lot of kind of patient portals fail in the past. So be interested to hear anything you have. Yeah so we you know it's a really important question.

[01:41:20] We have had some successes and some failures. When I first got to Verily, one of the things I was quite fascinated by with the baseline health study was that the most important part of it was actually the community that had been formed of the participants of the baseline health study who actually felt like this was their community.

[01:41:40] I mean, I'm sure that the participants also had other communities in life too, but they really felt like they owned being a part of this community, and therefore their participation in baseline health study and also their engagement was very correlated. We looked a lot at the relationship between

[01:42:00] between participant engagement and, for example, data quality. And what you can see is, and you wouldn't be surprised, the data quality is improved when people feel essentially directly responsible and also that there's benefit. But the other part was asking people questions about what is and is not okay.

[01:42:20] And we published in PLOS ONE about two years ago. Like lessons learned, one of them was that patients really did not want us to return their genomic results to their doctors until they had a chance to think about it and review it. So the rate was much higher than what we expected. And so the other

[01:42:40] used to that was ask the question. So I think that, thank you guys very, very much. Thank you. Good point. I have good news and bad news. First, one of the good news is we've arranged for a

[01:43:00] cocktail reception on the lawn overlooking the ocean. The bad news is that it's a little chillier than we had predicted. AI doesn't do everything we hoped. So if you have a coat, bring it along and please enjoy the views and the alcohol.

[01:43:20] all.